- 07 Feb, 2017 7 commits
-
-
Brian Christiansen authored
Continuations of 86ed603c CID: 161500, 45176
-
Tim Wickberg authored
-
Tim Wickberg authored
-
Tim Wickberg authored
No functional change.
-
Tim Wickberg authored
-
Dominik Bartkiewicz authored
Bug 3447
-
Brian Christiansen authored
Makes it so that an srun/salloc initiated from a cluster with a one switch type will work on a different cluster with a different switch type.
-
- 06 Feb, 2017 6 commits
-
-
Danny Auble authored
-
Danny Auble authored
-
Danny Auble authored
link to libslurm as well as libslurmdb to make things work.
-
Danny Auble authored
-
Danny Auble authored
dynamically or statically to libslurm. This can result in much smaller binaries, but isn't the easiest to develop against as if the files in src/common or src/*api change and the binaries linking against them are compiled directly libslurm.so doesn't get recompiled automatically, you have to manually do it. But in production this could give large benefits as Slurm's footprint is now considerably smaller. This used to not work on environments like AIX where plugins couldn't resolve variables used in the parent program, but this appears to not be the case now we link to the "full" .so which exports everything as the .o does.
-
Danny Auble authored
-
- 04 Feb, 2017 1 commit
-
-
Tim Wickberg authored
Dead code, since xmalloc will abort() on OOM.
-
- 03 Feb, 2017 9 commits
-
-
Alejandro Sanchez authored
Bug 3444
-
Morris Jette authored
-
Morris Jette authored
Remove a couple of unnecesaary fatal() calls and remove some other unnecessary logic.
-
Morris Jette authored
This would not really be a fatal condition (although we would eventually exhaust memory
-
Morris Jette authored
This would not really be a fatal condition
-
Morris Jette authored
bit_alloc() will abort if the malloc fails
-
Morris Jette authored
-
Morris Jette authored
All of these fatal conditions are already handled by xmalloc()
-
Morris Jette authored
-
- 02 Feb, 2017 17 commits
-
-
Morris Jette authored
These were false positive error reports, but should go away now.
-
Morris Jette authored
-
Morris Jette authored
These are all false positives, but code will remove the Coverity error reports
-
Morris Jette authored
Add upper bounds to elimate "Untrusted loop bound" errors.
-
Brian Christiansen authored
-
Brian Christiansen authored
Passes the client's origin cluster name to the controller. If the client's cluster name is different from the controller's cluster name it will send back the cluster and node's information for the salloc/srun to talk to. e.g. cluster1$ srun -Mcluster2 hostname
-
Brian Christiansen authored
-
Brian Christiansen authored
-
Brian Christiansen authored
if present.
-
Brian Christiansen authored
Previously it was just putting the IP and the srun would check to see if slurm.conf's cluster name and the IP were the same to prevent setting the working cluster unless if needed to. But since it was was checking the cluster name and the IP it would always try to set the working cluster rec even if they were the same.
-
Morris Jette authored
See if this tactic can be used to make a bunch of Coverity errors go away.
-
Morris Jette authored
Add NO_VAL32 (same value as NO_VAL) and add INFINITE32 (same value as INFINITE) These now match the NO_VAL16, NO_VAL64, INFINITE16 and INFINITE64 value types
-
Morris Jette authored
Coverity does not recongize the getopt_long() function's required argument, so Coverity reported errors whenever a NULL optarg value resulted in an error. This change adds a bunch of tests for NULL optarg value and skips it's processing.
-
Tim Shaw authored
Bug 3406.
-
Tim Shaw authored
Bug 3122.
-
Tim Shaw authored
In preparation for Bug 3122
-
Morris Jette authored
-